FSP's quiet work: capacitors, radiators, controllers

What the work claims

This is a primary program-status presentation by the NASA Glenn fission surface power team, not a peer-reviewed result paper. Its claims are claims of progress on maturation activities as of mid-2022. Four externally funded SBIR efforts are reported with bench outcomes: an oscillating heat pipe radiator panel from ThermAvant Technologies that rejected more than 800 W on a 0.5 square meter panel at a conductance of 80 W per kelvin against a 110 C heat source; a process from Peregrine Falcon Corporation for metallurgically joining Haynes 230 alloy to depleted-uranium surrogate core material by axial force; a Creare paper design of a 500 kWe supercritical Brayton converter, with Phase II funding directed at fabricating a 500 kWe turbomachine assembly; and Sigma Technologies polymer-multilayer (NanoLam) capacitors demonstrating 6.3 J per cubic centimeter in a 140 microfarad, 2200 V medical application, described as roughly a 50-fold energy-density improvement over MIL-PRF-39022/12 devices.¹

In-house, the Glenn team reports an integrated sodium heat pipe welded into a Stirling convertor heater head as a deliberate departure from the clamped-plate interface used on the KRUSTY reactor experiment; a simplified average-amplitude control approach for Stirling engines backed by a published 1 kW design; and a reconfigurable simulator rig, built on a dSPACE MicroLabBox, that emulates convertor alternator frequency response up to 1 MHz and can host up to four engines at once.¹ Every one of these is bench or desk work. None of it is flight hardware, and the presentation does not claim otherwise.

How it works

The thermal line is the most physical. A fission surface power system rejects its waste heat through deployable radiator panels, and the radiator is a large fraction of system mass. The ThermAvant SBIR tested an oscillating heat pipe (OHP) panel: a sealed plate with partially filled working fluid whose vapor bubbles and liquid slugs oscillate passively, moving heat with no pump. Against a 110 C heat-pipe-fluids-loop source the panel held an 80 W/K conductance and rejected more than 800 W from a 0.5 square meter panel; the stated Phase II objective is to optimize the same panel class for 175 C operation, with working-fluid screening across 100 to 300 C.¹ The maturity boundary matters: this is a single-panel bench test at 110 C, not a full radiator at the hot end of the fission loop temperature range.

The fuel-core joining work addresses a quieter problem. Heat pipe reactors depend on low-loss heat transfer from the reactor core into the heat pipes, and the joint between the Haynes 230 heat pipe wall and the uranium-molybdenum core is where that loss lives. Phase I demonstrated material compatibility and a strong metallurgical joint on flat coupons; Phase II moved to joining cylindrical coupons with axial force and measuring the shear strength of the consolidated joint.¹ Again the boundary is explicit: surrogate material, coupon scale, no irradiated hardware.

The electronics items look small but carry system weight. Stirling convertors need controllers because the thermal power arriving at the engine is effectively constant while the electrical extraction must be throttled: if energy is not drawn off, it accumulates in the piston motion and the convertor overstrokes. The team's bet is that historically complex digital control can be replaced by simplified average-amplitude control, cutting component count and hence validation burden and failure modes; a 1 kW capacitor-based power-factor-correction controller design was published at NETS 2021.¹ The simulator rig exists because controller development on real convertors risks damaging engines: it linearizes the convertor model in a dSPACE unit, reproduces the alternator's frequency response to 1 MHz, senses plus or minus 25 A per channel, and drives up to four engines from Chroma two-quadrant supplies.¹

The capacitor line has the hardest numbers. Dynamic radioisotope and Stirling power electronics need DC-link and power-factor-correction capacitors that survive radiation and high temperature without the bias-dependent degradation and fragility of ceramics. Sigma's NanoLam polymer-multilayer parts deliver 6.3 J/cc in a demonstrated 140 microfarad, 2200 V medical part, hold their properties to 200 C, self-heal into a fail-open mode, and were carried into FSP-relevant parts by the SBIR: a 50 Vdc, 750 microfarad DC link and, as a proof of concept, a 250 Vac, 80 microfarad AC power-factor-correction capacitor, with breakdown voltage at a 2.6 times safety factor and 900-hour short-term life tests complete.¹

The Brayton item is the least mature: Creare's Phase I was explicitly a paper design study of a 500 kWe supercritical Brayton converter, trading working fluid, turbine inlet temperature, pressure ratio, recuperator effectiveness, and alternator cooling, with Phase II funding aimed at fabricating a 500 kWe turbomachine assembly. That is a component with no rotating hardware demonstrated yet under this contract.¹

The strongest case for it

The steelman is that this is what a serious fission program looks like from the inside. Reactors attract the attention, but the integration failures that kill schedules hide in passives, joints, and controllers. A capacitor that stores 50 times the energy per volume of the mil-spec baseline directly shrinks the DC-link and power-processing hardware that Stirling-based systems carry on every string; the presentation notes the same SBIR fed dynamic radioisotope power systems, so one component effort serves two program lines.¹ A welded heat pipe to heater-head joint, if it survives vacuum extended operation as planned, removes a thermal interface that the KRUSTY-era clamped design accepted as a loss source. And the simulator rig is genuinely good engineering practice: validating controllers against an emulated fleet of engines before touching hardware is exactly how a program with a 10-year no-maintenance reliability requirement should spend its money.¹

Where a skeptic should push

Four pressure points. First, every maturity claim here is bench-scale and the deck is a status snapshot: the OHP panel was tested at 110 C with a Phase II goal of 175 C, so the fission-relevant operating point was not yet demonstrated; the capacitors carry 900-hour life tests against a 10-year mission requirement, four orders of magnitude in time; the DU joining is on surrogate coupons; the Brayton converter is a paper study. Second, SBIR attrition is real and the deck reports program structure, not downselection outcomes: nothing here proves these efforts survived contact with the industry-led system design, which the presentation itself notes belongs to the contractors.¹ Third, the 6.3 J/cc figure comes from a medical-application part, not an FSP flight part; the FSP-relevant outcomes are the 50 V and 250 V proof-of-concept units, whose energy density the deck does not state. Treat 6.3 J/cc as a technology ceiling from a different application, not as a delivered FSP spec. Fourth, this record is from August 2022; program direction has shifted since (the same NTRS series now includes a 2025 instrumentation-and-controls maturation webinar), so none of these lines should be read as current status without a fresh primary record.¹

What FSP maturation means for space power

The non-obvious implication: for space fission, the critical path runs through the least glamorous components. A 40 kW lunar reactor needs Stirling engines, and Stirling engines need DC-link capacitors and fault-tolerant controllers; the deck shows these items still at SBIR and breadboard scale even as the system design was being competitively procured, which suggests the binding constraint on lunar fission deployment is not reactor physics but the radiation-and-temperature-qualified passives and control electronics supply chain.¹ The genuine threat is qualification time: 900-hour capacitor life tests against a 10-year, zero-maintenance requirement mean a long, expensive reliability tail, and a single passive class failing qual can idle an entire reactor program the way connector and bearing problems have idled propulsion programs.¹ The opportunity is architectural: if the OHP radiator panel hits its 175 C Phase II target and the welded heat pipe joint holds up in vacuum, fission surface power systems shed both pumping parasitics and a thermal interface loss, margin that flows directly into the mass budget that surface power never has enough of.¹

The bottom line

Calibrated judgment, high confidence in what the document says, low confidence in where these lines stand today. As of August 2022, NASA's fission surface power program was doing the unglamorous component maturation that serious space power programs require, with specific, checkable bench results: an 800 W-class OHP panel at 80 W/K, surrogate-core heat pipe joints, a four-engine Stirling simulator, and polymer capacitors at roughly 50 times mil-spec energy density in a non-space part. What would confirm the thesis: a current primary record showing the 175 C OHP panel tested, FSP-grade capacitors past 10-year-equivalent qualification, or a welded heat pipe convertor through thermal-vac. What would break it: downselection records showing these SBIR lines were dropped when industry took over system design.